Extended Infusion of Beta-Lactams and Glycopeptides: A New Era in Pediatric Care? A Systematic Review and Meta-Analysis

Optimizing antibiotic therapy is imperative with rising bacterial resistance and high infection mortality. Extended infusion defined as a continuous infusion (COI) or prolonged infusion (PI) of beta-lactams and glycopeptides might improve efficacy and safety compared to their intermittent administration (IA). This study aimed to evaluate the efficacy and safety of extended infusion in pediatric patients. Adhering to Cochrane standards, we conducted a systematic review with meta-analysis investigating the efficacy and safety of COI (24 h/d) and PI (>1 h/dose) compared to IA (≤1 h/dose) of beta-lactams and glycopeptides in pediatrics. Primary outcomes included mortality, clinical success, and microbiological eradication. Five studies could be included for the outcome mortality, investigating meropenem, piperacillin/tazobactam, cefepime, or combinations of these. The pooled relative risk estimate was 0.48 (95% CI 0.26–0.89, p = 0.02). No significant differences between the administration modes were found for the outcomes of clinical success, microbiological eradication (beta-lactams; glycopeptides), and mortality (glycopeptides). No study reported additional safety issues, e.g., adverse drug reactions when using COI/PI vs. IA. Our findings suggest that the administration of beta-lactams by extended infusion leads to a reduction in mortality for pediatric patients.


Introduction
The global rise in antibiotic resistance due to misuse and overuse of antibiotics [1] complicates infection treatment, potentially leading to treatment failure [2][3][4][5][6], while the development of new antibiotics remains limited [7].In pediatrics, restricted antibiotic options, with contraindications for tetracyclines and fluoroquinolones [6,8], contribute to the challenge.Approximately 37% to 61% of hospitalized pediatric patients receive antibiotics, making them the most prescribed drugs in pediatrics [9].WHO reports that multidrug-resistant bacterial infections cause 700,000 global fatalities annually, including 200,000 newborns [6].Optimizing antibiotic treatment is crucial for improving efficacy and safety, with rapid detection of bacterial infection and the selection of effective and safe treatments being essential for reducing mortality [10][11][12].

Antibiotic Resistance
The Swiss Strategy against Antibiotic Resistance (StAR) emphasizes the importance of proper antibacterial agent use to mitigate the development of antibiotic resistance and prevent infections with resistant pathogens [1].
Appropriate usage should align with pharmacokinetic (PK) and pharmacodynamic (PD) data for efficacious pathogen eradication [13].Consequently, hospitals globally have implemented antimicrobial stewardship programs to advocate for judicious antibiotic usage [3].

PK/PD of Beta-Lactams and Glycopeptides
Beta-lactams and glycopeptides, the predominant antibiotics in pediatrics, are timedependent bactericidal drugs that inhibit bacterial cell wall synthesis [14,15].Their effectiveness depends on substantiating the free drug concentration above the minimum inhibitory concentration (MIC) for bacterial growth [15,16].
The MIC value varies depending on the pathogen and the antibiotic utilized [15].It is crucial to maximize the fraction of time (fT) that the antibiotic concentration remains above the MIC.The minimum efficacy targets for pediatrics include fTs of >50% for penicillins, >60% for cephalosporins, and >40% for carbapenems [15].The efficacy further increases when concentrations exceed the MIC, extending up to four times the MIC for beta-lactams [17].For glycopeptides, the therapeutic target is an fT > MIC of 100% [15].These objectives are frequently not achieved with the standard intravenous (i.v.) intermittent administration (IA), but they may be attainable with continuous infusion (COI) [16].As illustrated in Figure 1, plasma concentration fluctuates with IA but remains constant with COI (own illustration).Furthermore, COI may diminish the emergence of antibiotic resistance [18], decrease the risk of inadequate antibiotic concentration, and enhance the efficacy of infection treatment [8,19].

Antibiotic Resistance
The Swiss Strategy against Antibiotic Resistance (StAR) emphasizes the importance of proper antibacterial agent use to mitigate the development of antibiotic resistance and prevent infections with resistant pathogens [1].
Appropriate usage should align with pharmacokinetic (PK) and pharmacodynamic (PD) data for efficacious pathogen eradication [13].Consequently, hospitals globally have implemented antimicrobial stewardship programs to advocate for judicious antibiotic usage [3].

PK/PD of Beta-Lactams and Glycopeptides
Beta-lactams and glycopeptides, the predominant antibiotics in pediatrics, are timedependent bactericidal drugs that inhibit bacterial cell wall synthesis [14,15].Their effectiveness depends on substantiating the free drug concentration above the minimum inhibitory concentration (MIC) for bacterial growth [15,16].
The MIC value varies depending on the pathogen and the antibiotic utilized [15].It is crucial to maximize the fraction of time (fT) that the antibiotic concentration remains above the MIC.The minimum efficacy targets for pediatrics include fTs of >50% for penicillins, >60% for cephalosporins, and >40% for carbapenems [15].The efficacy further increases when concentrations exceed the MIC, extending up to four times the MIC for beta-lactams [17].For glycopeptides, the therapeutic target is an fT > MIC of 100% [15].These objectives are frequently not achieved with the standard intravenous (i.v.) intermittent administration (IA), but they may be attainable with continuous infusion (COI) [16].As illustrated in Figure 1, plasma concentration fluctuates with IA but remains constant with COI (own illustration).Furthermore, COI may diminish the emergence of antibiotic resistance [18], decrease the risk of inadequate antibiotic concentration, and enhance the efficacy of infection treatment [8,19].The time-dependent pharmacokinetic/pharmacodynamic (PK/PD) mode of action of beta-lactams and glycopeptides suggests that COI might be less toxic and more effective than IA [19,20].By minimizing peak concentrations, both COI and prolonged infusion (PI) may result in fewer adverse drug reactions (ADRs), such as nephrotoxicity, hypersensitivity reactions, or neurological deterioration [21,22].
COI also facilitates therapeutic drug monitoring (TDM) by maintaining a constant drug concentration (Figure 1B).In contrast to IA, the timing of blood sampling and antibiotic dose administration is not critical.This reduces the risk of incorrect dose adjustments [12,14].When employing COI, the area under the curve (AUC), which is in this case used for dose adjustments, can be easily calculated by multiplying the drug The time-dependent pharmacokinetic/pharmacodynamic (PK/PD) mode of action of beta-lactams and glycopeptides suggests that COI might be less toxic and more effective than IA [19,20].By minimizing peak concentrations, both COI and prolonged infusion (PI) may result in fewer adverse drug reactions (ADRs), such as nephrotoxicity, hypersensitivity reactions, or neurological deterioration [21,22].
COI also facilitates therapeutic drug monitoring (TDM) by maintaining a constant drug concentration (Figure 1B).In contrast to IA, the timing of blood sampling and antibiotic dose administration is not critical.This reduces the risk of incorrect dose adjustments [12,14].When employing COI, the area under the curve (AUC), which is in this case used for dose adjustments, can be easily calculated by multiplying the drug concentration with the duration of application.Through COI, the attainment of the target concentration is achieved more efficiently for both glycopeptides [19] and beta-lactams [23], potentially improving clinical outcomes [24].
Evidence increasingly supports the correlation between meeting PK/PD targets and achieving clinical success in adults [12,25,26].Although PK studies in pediatric patients suggest potential advantages of COI/PI over IA [27][28][29][30], there is limited evidence establishing its superiority in clinical efficacy and safety to warrant its adoption as the standard method for pediatric patients [8,16,31].

Considerations for the Pediatric Population
Developmental stages significantly alter PK (see Table 1), causing variations in antibiotic plasma concentrations based on changes in distribution volume, blood clearance, and drug half-life [31].Primary factors contributing to this variability include body weight and maturation effects [27].Despite demonstrated clinical benefits of COI/PI for betalactams and glycopeptides in adults [7,11,32], we cannot extrapolate the same outcomes for pediatric patients.SwissPedDose guidelines currently recommend IA for almost all beta-lactams and glycopeptides, with PI limited to meropenem and ceftazidime specifically for cystic fibrosis [33].Notably, there is no recommendation for COI/PI for glycopeptides in SwissPedDose as of now [33].

Objective
We hypothesized that COI of beta-lactams and glycopeptides would be effective and safe in pediatrics.Our objective was to perform a systematic review with a meta-analysis to evaluate the safety and efficacy of COI/PI of beta-lactams and glycopeptides compared to IA in pediatrics, consolidating existing knowledge for a robust conclusion [35].

Study Selection
The initial search, limited to RCTs, yielded 581 studies, with four deemed relevant for full-text screening.Following the protocol, the search was expanded to include observational studies [36].Across EMBASE, CENTRAL, and PUBMED, we identified 3066 studies.After duplicate removal (n = 422) and title and abstract selection, we found nine studies to be included in the meta-analysis during the full-text analysis-six on beta-lactams and three on glycopeptides (Figure 2).

Risk of Bias Assessment
With the exception of one study, the overall risk of bias (ROB) assessment was moderate or high, as indicated in Tables 2 and 3, with additional details provided in Supplement F.
Table 2. Overall risk of bias of primary outcomes in studies investigating beta-lactams.

Mortality
Clinical Success Microbiological Eradication Beauchamp 2019 [37] High risk High risk High risk Chongcharoenyanon 2021 [17] Some concerns High risk n/a Shabaan 2017 [21] Low risk Low risk Low risk Solorzano 2019 [38] High risk High risk n/a Zembles 2021 [39] Some concerns Some concerns n/a Zembles 2022 [40] Some concerns Some concerns n/a n/a not applicable.
Table 3. Overall risk of bias of primary outcomes in studies investigating glycopeptides.

Mortality
Clinical Success Microbiological Eradication Demirel 2015 [41] High risk High risk High risk Gwee 2015 [42] Some concerns n/a n/a Wysocki 2022 [43] High risk High risk High risk n/a not applicable.
The Wysocki study was excluded from data synthesis due to its study design, which could introduce significant bias to our outcome analysis (Supplement F).Sensitivity analysis showed no substantial modification of effect estimates after excluding studies classified as having a high ROB (Supplement G).

Risk of Bias Assessment
With the exception of one study, the overall risk of bias (ROB) assessment was moderate or high, as indicated in Tables 2 and 3, with additional details provided in Supplement F.
Table 2. Overall risk of bias of primary outcomes in studies investigating beta-lactams.

Mortality Clinical Success Microbiological Eradication
Beauchamp 2019 [37] High risk High risk High risk Chongcharoenyanon 2021 [17] Some concerns High risk n/a Shabaan 2017 [21] Low risk Low risk Low risk Solorzano 2019 [38] High risk High risk n/a Zembles 2021 [39] Some concerns Some concerns n/a Zembles 2022 [40] Some concerns Some concerns n/a n/a not applicable.
Table 3. Overall risk of bias of primary outcomes in studies investigating glycopeptides.

Mortality Clinical Success Microbiological Eradication
Demirel 2015 [41] High risk High risk High risk Gwee 2015 [42] Some concerns n/a n/a Wysocki 2022 [43] High risk High risk High risk n/a not applicable.
The Wysocki study was excluded from data synthesis due to its study design, which could introduce significant bias to our outcome analysis (Supplement F).Sensitivity analysis showed no substantial modification of effect estimates after excluding studies classified as having a high ROB (Supplement G).

Study Characteristics
Study characteristics are outlined in Tables 4 and 5, with additional details available in Supplement H.The administered drug doses in all studies were within the recommended range of SwissPedDose [33].Abbreviations: AB = antibiotic; CS = cohort study; FEP = cefepime; MEM = meropenem; RCT = randomized controlled trial; TZP = piperacillin and tazobactam.Abbreviations: AB = antibiotic; COI = continuous infusion; EMR = electronic medical records; RCT = randomized controlled trial; VAN = vancomycin.

Beta-Lactams: Outcome Mortality
In the COI/PI group, 2.8% (13/458) of patients died, compared to 5.6% (32/567) in IA group [21,[37][38][39][40].The pooled RR estimate was statistically significant (RR = 0.48; C 0.26 to 0.89; p = 0.02).Except for one study, all contributing studies included the no-eff value in the CI, consistent with the prediction interval.Tau 2 and I 2 statistics were 0 a 0%, respectively, as Chi 2 was smaller than the degrees of freedom (df).Visual inspect of the forest plot suggested high heterogeneity between studies (Figure 3). Figure 3 presents the mortality outcomes associated with beta-lactam antibiotics i forest plot.Statistically non-significant effect estimates of all outcomes and repor adverse drug reactions (ADRs) for both antibiotic groups are detailed in Supplement I

Subgroup Analyses
Subgroup analyses, as planned in the protocol (Supplement J), could not conducted due to the lack of studies [36].

Publication Bias
None of the outcomes had a sufficient number of studies (n ≥ 10) for a meaning assessment of publication bias using funnel plots (Supplement K).As a result, detectable publication bias was observed.

Study Total (95% CI) Prediction interval
Figure 3 presents the mortality outcomes associated with beta-lactam antibiotics in a forest plot.Statistically non-significant effect estimates of all outcomes and reported adverse drug reactions (ADRs) for both antibiotic groups are detailed in Supplement I.

Subgroup Analyses
Subgroup analyses, as planned in the protocol (Supplement J), could not be conducted due to the lack of studies [36].

Publication Bias
None of the outcomes had a sufficient number of studies (n ≥ 10) for a meaningful assessment of publication bias using funnel plots (Supplement K).As a result, no detectable publication bias was observed.

Certainty of Evidence
We assigned a "very low" certainty of evidence for primary outcomes in both drug groups (Table 6).Abbreviations: AB = antibiotic; BL = beta-lactam; CI = confidence interval; COI = continuous infusion; IA = intermittent administration; GP = glycopeptide; GRADE = grading of recommendations assessment, development, and evaluation; N = number of studies; nRCT = not exclusively RCTs; RCT = randomized controlled trial; ROB = risk of bias; RR = relative risk (risk ratio). 1 Single study, inconsistency not applicable (n/a).

Overview of Findings
To our knowledge, this is the first meta-analysis comparing extended infusion with IA for glycopeptides in pediatric patients and one of the first ones for beta-lactams.Our results indicate a statistically significant lower pooled RR for mortality when administering beta-lactams via extended infusion instead of IA.This aligns with findings from adult studies, where optimal antimicrobial exposure is associated with better clinical outcomes and lower mortality [11,25].Additionally, also very recently published data from the metaanalysis of Budai et al. align with our results [44].Several Monte-Carlo simulation studies investigating the probability of PK/PD target attainment in pediatrics also support the use of COI for beta-lactams, further validating our findings [27,45].Consistent with the findings of Grupper et al. [46], we did not observe COI/PI to be inferior to IA in terms of the safety and efficacy of glycopeptides.However, we did not find significant beneficial differences in mortality, clinical success, or microbiological eradication when using glycopeptides.
Potential reasons for this include the study setting, aiming for high target concentrations to combat even the most resistant pathogens.If patients had infections caused by pathogens requiring lower antibiotic concentrations, IA possibly remained sufficiently effective.Consequently, any potential superiority of COI/PI would not have been observable.Supporting this finding, clinical trials showed that clinical success was still achieved, event when most patients did not reach the target antibiotic blood concentrations [25,47].Therefore, non-critically ill patients may not experience the same clinical benefits as critically ill patients, who typically require higher antibiotic efficacy to combat pathogens [48].This aligns with research on COI in adults and echoes clinical suggestions applied in the clinic of Heidenheim [25,39,47,49,50].Additionally, this assumption is supported by Shabaan et al., the only beta-lactam study included in the meta-analysis conducted on critically ill patients [21].It was the only study to demonstrate statistically significant benefits for all three primary outcomes assessed.While direct evidence is lacking, the literature indirectly suggests that COI may reduce antibiotic resistance emergence by avoiding subtherapeutic concentrations [11,51].However, this hypothesis requires further investigation.

Implications for Clinical Practice and Outlook
COI of beta-lactams and glycopeptides aligns better with the PK/PD profile than IA [49,52].Our findings, demonstrating safety and efficacy benefits in pediatric patients, are supported by a review on COI use in the pediatric population [7,49,53,54].Cheng et al. also reported that COI of meropenem was more effective in treating sepsis than IA [55].COI offers additional potential benefits, such as reducing patients' length of stay and enhancing cost-effectiveness, which can be attributed to increased therapy efficacy and align with the interests of hospitals [8,11].
We recommend conducting an RCT to further explore the comparative benefit of COI combined with therapeutic drug monitoring (TDM) versus IA for the specified beta-lactams and glycopeptides in our guideline.

Strengths and Limitations
Our systematic review employed well-documented methods outlined in a registered protocol, adhering to the Cochrane Handbook of Systematic Reviews, with a specific focus on the relatively unexplored pediatric population [36,56].
A primary limitation was the scarcity of studies available for inclusion in the metaanalysis.The identification of only six eligible beta-lactam studies (including three RCTs) and three glycopeptide studies (including one RCT) underscores the reported lack of evidence [49].Ethical considerations and insufficient financial incentives for conducting such studies in pediatrics may contribute to the limited research.With over 50% of the studies being non-RCTs, uncontrolled confounding factors could have influenced our results.The shortage of studies constrained both publication bias assessment and subgroup analysis, necessitating the pooling of all pediatric age groups, beta-lactam drugs, COI with PI, and all disease severities.This, in turn, compromised the robustness of the pooled RR estimates.Exclusion of the Shabaan et al. study, which showed reduced mortality with COI for beta-lactams [21], would have prevented achieving statistical significance.This study had the highest weight (inverse variance), further reducing the overall robustness.However, it was the only study with a low risk of bias, enhancing the credibility of the effect estimate.To elevate the low GRADE of the findings, additional well-designed RCTs are imperative to further explore the benefits of COI.
A limitation is excluding patients on continuous renal replacement therapy (CRRT) to minimize confounding, despite our advocacy for COI for critically ill patients, many of whom require CRRT.Future research should explore CRRT's impact on outcomes of COI of antimicrobials.Additionally, our study did not consider pharmacokinetic changes like fluid shifts or organ dysfunction associated with severe illness [57].We were also unable to adjust for concomitant antibiotic use such as aminoglycosides, potentially confounding our findings.

Conclusions
Our research suggests that COI of beta-lactams and glycopeptides for pediatric patients is feasible, safe, and more efficacious.Existing PK simulation studies and those conducted with adults support the benefits of COI.However, further validation of our findings and paving the way for clinical implementation require more RCTs.Therefore, we propose conducting an RCT to investigate the comparative benefits of COI combined with TDM over IA of beta-lactams and glycopeptides, specifically in critically ill pediatric patients who are likely to derive the most benefit from COI.

Eligibility Criteria
Following the Cochrane Handbook of Systematic Reviews, we formulated a study protocol adhering to the PRISMA-P checklist to ensure transparency and comprehensiveness [56,58].The protocol was registered in PROSPERO (CRD42023407772) before the literature search [36].Deviations from the protocol were documented (Supplement A).We established inclusion and exclusion criteria for the publications (Table 7).For infusion durations, we defined 24 h/day as COI, ≤1 h as IA, and >1 h <24 h/day as PI.• IA in the context of CRRT (a potential source of bias, altering PK [59,60]).Abbreviations: COI = continuous infusion, CRRT = continuous renal replacement therapy, IA = intermittent administration, i.v.= intravenous, PI = prolonged infusion, PK = pharmacokinetics, and RCT = randomized controlled trial.

Search Strategy and Information Sources
After refining the search strategy with a professional librarian [61], we systematically searched EMBASE, MEDLINE, and CENTRAL databases for relevant studies published between 1960 and 17 April 2023 (Supplements B and C).To filter for randomized controlled trials (RCTs), we utilized the Cochrane filter [62].A second search, including observational studies, was conducted, as the initial search yielded fewer than 10 RCTs, as predefined in the protocol [36].No other search filters were applied.We validated the search by checking the inclusion of defined key papers.Before data synthesis, we reran the search to include new publications.Duplicate removal was conducted manually using the deduplication tool in EndNote, comparing titles, years, and authors, followed by the digital object identifier (DOI) if available [63].Additional relevant publications were manually sought in the references of reviews and studies included in the full-text review.

Study Selection
Two independent reviewers screened titles and abstracts for inclusion in the full-text assessment, with a third reviewer resolving discrepancies.Full texts of eligible studies were screened based on inclusion and exclusion criteria (Table 7).Results were compared after each step, with consensus decisions in cases of differences.

Data Collection and Analysis
Following the Cochrane Handbook for Systematic Reviews, two independent reviewers conducted the data extraction in an EXCEL table [56,64].Applicability was tested using a sample study, and differences were resolved through consensus, involving a third reviewer when necessary.In cases of missing data, study authors were contacted via email for the required information.Relative risks (RRs) were chosen as the measure of effect for the outcomes (Supplement D).A statistically non-significant result on a p-level of 0.05 was indicated if the 95% confidence interval (CI) contained the value one.To prevent calculation errors due to division by zero in studies with no events in one arm, an event value of 0.5 was used, adhering to the Cochrane Handbook for Systematic Reviews of Interventions [56].Studies with no events in either study arm for a specific outcome were excluded from the meta-analysis.

Risk of Bias Assessment
Two independent reviewers used the Cochrane tool for randomized trials ROB2 [65] to assess the risk of bias (ROB) for each outcome.Non-randomized trials were evaluated using the risk of bias in non-randomized studies of interventions (ROBINS-I) tool [66].A sensitivity analysis was performed to gauge the robustness of the effect estimates by excluding studies with a high ROB.The reviewers reached a consensus to determine if the ROB for each study was too high for inclusion in the analysis.

Data Synthesis
Glycopeptides and beta-lactams data syntheses were conducted independently.A meta-analysis with a random effects model was performed for each outcome of each antibiotic group if at least two studies reported the outcome.Forest plots were generated using the meta package in R in RStudio [67,68], employing inverse variance to weight studies as per the Cochrane handbook [56].If fewer than two studies reported the outcome, results were reported in prose.

Subgroup Analyses and Heterogeneity Assessment
Planned subgroup analyses for both drug groups encompassed different drugs, age, sex, treatment indication, infectious agent, severity of infection, concomitant diseases, and concurrent use of other antibiotics [36].Forest plots were utilized for qualitative heterogeneity assessment, where a significant overlap of CI indicated high heterogeneity [56].Qualitative and quantitative heterogeneity assessment employed Chi 2 tests, the Higgins I 2 statistic, τ 2 , and prediction intervals calculated in R using the meta and metafor packages [56,67,69,70].To address heterogeneity resulting from pooling PI with COI and pooling different beta-lactam antibiotics, a random effects model was chosen.

Publication Bias
We planned to assess publication bias by visually inspecting funnel plots of primary outcomes, created using the metafor package in R [68,69].A minimum of 10 studies is generally considered sufficient for adequate test power to assess funnel plot asymmetry [56,71].If publication bias was detectable, we intended to use the trim and fill function to obtain an effect estimate of the true unbiased effect [56].

Grade Assessment
We used the grading of recommendations assessment, development, and evaluation (GRADE) approach to assess the certainty of the evidence for each primary outcome [56,72].

Figure 1 .
Figure 1.Comparison of the fraction of time (fT) that the plasma concentration remains above the minimal inhibitory concentration (MIC) in intermittent administration (A) and continuous infusion (B).In intermittent administration, there are efficacy gaps where the fT < MIC (light red).With continuous infusion, a constant fT of 100% > MIC (light green) can be maintained throughout the entire administration period.

Figure 1 .
Figure 1.Comparison of the fraction of time (fT) that the plasma concentration remains above the minimal inhibitory concentration (MIC) in intermittent administration (A) and continuous infusion (B).In intermittent administration, there are efficacy gaps where the fT < MIC (light red).With continuous infusion, a constant fT of 100% > MIC (light green) can be maintained throughout the entire administration period.

Figure 2
presents the results of the mentioned selection process in a flow diagram.Further exclusion details are available in Supplement E.

Figure 2 .
Figure 2. Flow diagram of the selection process of studies for inclusion in the meta-analysis.

Figure 2
Figure 2 presents the results of the mentioned selection process in a flow diagram.Further exclusion details are available in Supplement E.

Figure 2 .
Figure 2. Flow diagram of the selection process of studies for inclusion in the meta-analysis.

Table 4 .
Characteristics of beta-lactam studies included in the systematic review.

Table 5 .
Characteristics of glycopeptide studies included in the systematic review.

Table 6 .
GRADE evidence profile: continuous infusion of beta-lactams and glycopeptides in pediatric patients.

Table 7 .
Inclusion and exclusion criteria for studies for the systematic review.